| the Bay Area experience |
[08 Jul 2009|09:20pm] |
Today I met a cross-section of the "creative class"... which was fun and interesting. It's also interesting to look at the culture that arises in such environments. Our highest value is openness #opendata. Not only do we want data to be free, but we want to produce lots of it, rather indiscriminately. It's no surprise that Twitter is their icon... and we are living up to Herb Simon's prediction. Although there's an element of trendiness and geek-chic to this, most people seem to be the real thing.
Experiencing this sort of thing was one of my main reasons for visiting the Bay Area.
Also: I got WiFi at the BART, and in a moving car in downtown Palo Alto (though neither worked very well)
The whole idea of an unconference is that it's a dynamic event, in which the audience decides which events take place. Today this involved making proposals in A5-sized pieces of paper containing a number of bubbles, and "voting" by filling in a bubble. This information is used for room assignment.
Here's something about the people I met today:
* I met at least 3 transhumanist types
* One lady works for Nature Magazine, Second Life division (she's just 1 out of 2.5).
* Dawei Lin, Director of Bioinformatics Core at UC Davis, has made Lego models of a virus at the Maker Faire
* 23andMe, a personal genomics company, has a representative
* PLoS has a representative
* Melanie Swan
* Tantek, who is just "t" on Twitter, and gets annoyed at n00b twitterers (migrating from MySpace) who write "at" as "@t".
* Naomi Most, science radio show host, who took my picture with ![[info]](http://l-stat.livejournal.com/img/userinfo.gif) simonfunk
|
|
| SciBarCamp |
[08 Jul 2009|02:40pm] |
I'm at SciBarCamp, my first unconference. Lots of compbio!
Dinner at Miyake. Added 3 people on twitter, many of whom have been tweeting as it happens.
http://twitpic.com/9qlwj
|
|
| Skype credit expiration |
[06 Jul 2009|07:02pm] |
=== What are the rules for Skype Credit expiry? ===
1. Skype Credit expires 180 days after your last credit purchase or action that used credit eg calling landlines or cell phones, sending an SMS message.
2. Each purchase, call or SMS message resets the expiry time to 180 days.
3. Unfortunately, if you don't use your remaining credit, it will expire to comply with normal business accounting rules.
4. We don’t want you to lose your credit, so we send reminder emails 30 days, 7 days and 72 hours before your credit expires.
I propose a website to help you get around Skype's credit expiration policy: every 6 months, it logs you in to your account, and makes a 1-second phone call.
"Normal business accounting rules" sounds like BS.
|
|
| bleg: Facebook - local friend search |
[05 Jul 2009|05:56pm] |
I'd like to search for Facebook friends who are in the Bay Area.
I'd settle for friends who list themselves as in "San Francisco", etc. But if I type this in the search for people, it finds people who are named "San Francisco".
|
|
| hello San Francisco |
[04 Jul 2009|08:29pm] |
rdore is having a party. rjmccall told me about the urban hike yesterday (with a ridiculous number of CMU folks). I'd be interested in doing something like that in the next few days.
My phone number is 505-316-4892.
|
|
| Seattle for 1 night |
[02 Jul 2009|12:56am] |
|
I'm looking for a place to crash in Seattle on the night of the 13th. To repeat my Facebook offer: I will make a $25 Kiva loan if I get a couch, or $50 for a mattress.
|
|
| Nobel gossip |
[01 Jul 2009|12:53pm] |
I just had a 5-minute chat with Murray Gell-Mann over lunch. (LOLs omitted)
G - did you start this place?
M - yeah, it was mostly me.
...
M - universities can't do what we do. They have departments, grants, etc.
G - I was at CMU the last two years, and they were pretty good about interdisciplinarity.
M - CMU? what?
G - Cárnegie Mellon.
M - ah, Car-nay-gie Mellon.
M - I know Herb Simon. Herb was a good guy.
M - once we were sitting in the Presidential Science Committee(?), and Herb slipped me a note:
M - it said: "Murray, help me destroy the humanities!"
G - he was only half-joking, then?
M - he wasn't joking at all! He hated them.
...
G - I've read Feynman's book: "Surely you're joking"
M - Nonsense! Feynman spent huge amounts of time creating stories about himself, so he could write about them later.
M - One time he decided to never brush his teeth again.
M - He had horrible teeth.
M - We had the same dentist, and the dentist would bring him books and articles, but he never convinced Feynman.
M - He thought there was no proof that brushing your teeth did anything.
M - He told me: "Murray, you're a very conventional person, you're like a salesman... you wash your hands after going to the bathroom before you go to lunch! You only need to wash your hands if you pee on them."
M - in the Faculty Lounge, the men had to wear suits. When it was lunchtime, Feynman would take off his suit and tie, and hang it up in his office...
When he arrived (without a suit&tie), he would go to the cloakroom where they had this "public" suit and tie, for the people who forgot to bring theirs... and the suit was old and scratched, and the tie was bright orange.
G - he sure enjoyed the attention, huh?
UPDATE: this interview with Feynman confirms his skepticism about tooth-brushing.
|
|
| personal genomics; eye color; sexual selection: cognitive vs genetic |
[30 Jun 2009|09:55pm] |
If I get my DNA sequenced at 23andMe or DecodeMe, what information do I get? Is my privacy safe?
If my parents do it too, will it tell me where recombination happened in each chromosome? Btw, does anyone know of a visualization of the chromosomes showing some genes (and corresponding phenotypes)?
Kinda like this but more general-purpose and for laypeople.
---
Apparently, there is uncertainty about exactly which chromosomes contain genes related to eye color. GNXP.
As much as I dislike popsci, being a layman, this looks like the best explanation that I can understand (of the ones I could find through Google) of the fact that two blue-eyed parents can have a brown-eyed child: Eye color is a complex trait that depends on the state of several interacting genes. The gene that usually decides the issue (blue eyes or brown eyes) is the OCA2 gene on chromosome 15. But it comes in different strengths. A person with a weak form of the OCA2 gene will have blue eyes. Likewise a person with a strong form will have brown eyes.
The plot thickens, though, because an individual also has other eye-color genes that each has a say in the final eye-color outcome. For example, if one of these lesser genes is strong, it can make the weak form (blue) of OCA2 work much more effectively — almost like the strong form (brown). Then the eye color may be a light brown or muddy grey. In fact, the resulting color can be any shade of brown, hazel/green, or blue depending on the strengths of the interactions. Of course, there's mutation, but I'd expect that to be too rare to explain this phenomenon.
---
Now, blue-blue couples almost always have blue-eyed kids. This article is about the theory that blue-eyed men are more attracted to blue-eyed women because cheating would be easy to catch... of course, this wouldn't be that useful in populations where >%80 of the men are blue-eyed (and most of the others are heterozygotes).
So should we expect this effect to be stronger in more mixed populations? I imagine so; not because the blue-eyed men in those regions have stronger genes for blue-eye preference (this is bordering on silly), but because sexual attraction has a cognitive component. (The cognitive theory predicts adaptation within 1 generation; the natural selection theory would require many. But my main reason for believing the former is that the idea of such a specific gene sounds silly.)
In any case, this suggests that blue-eyed kids have enjoyed more paternal attention than brown-eyed kids.
The natural selection theory (though I'm not sure if anyone believes it) would be an instance of a common fallacy: "there's a gene for everything, no matter how specific".
|
|
| Berkeley / San Francisco |
[30 Jun 2009|09:40pm] |
I'm arriving at SFO on the 4th of July, ~4pm.
Which parties should I attend?
|
|
| batch video compression |
[30 Jun 2009|12:41am] |
I finally found a reasonable way to compress the AVI movies made by my Canon PowerShot. FormatFactory 1.90 works perfectly, and automatically converts multiple files, unlike Windows MovieMaker.
The GUI is confusing: you first need to add all files to the main window. Then you click on "Start".
I'm converting them to WMV (just for the heck of it). I've tried MP4 too, and it worked, with a similar compression rate/quality (~6 times smaller than the AVI, perfect quality). I've yet to try MPG.
If you're not careful, it will add a shortcut to eBay on your Desktop when you install it. In any case, that's a pretty small price to pay for effective freeware.
|
|
| my 3rd citation |
[25 Jun 2009|01:28am] |
|
This paper also cites my UAI paper (in its introduction, in its overview of LiNGAM). I don't know of any site that will keep an updated list of who's citing me.
|
|
| spam wave |
[24 Jun 2009|05:53pm] |
For the last week or so, a large wave of spam has been making it through my GMail filter. Like 5 / day, when previously it was like one a month (hard to verify, since GMail doesn't record which spam messages went to my Inbox originally).
I use several emails at optimizelife.com, and the spam I'm seeing comes from pretty much all of them. Conclusion: either the spammers have uniformly adapted the spam, or my GMail filter has become uniformly more lenient.
This probably isn't my first spam wave on GMail, but it's the most noticeable so far.
|
|
| the physical Church-Turing thesis |
[22 Jun 2009|08:54pm] |
Today, Cris Moore gave an excellent lecture about computational complexity. I managed to record some bits. He had a Scott Aaronson quote
Computers play the same role in complexity that clocks, trains, and elevators play in relativity. which sounds a lot like this Dijkstra one:Computer science is no more about computers than astronomy is about telescopes.
In an offline conversation afterwards, the topic of alternative computers (e.g. DNA computers, brains, etc) came up.
As we know, in terms of computability, the CT thesis says that all realizable computations can be simulated by a Turing Machine. Pretty much everybody believes this. But it says nothing about complexity (whether time or space). Occasionally people claim that alternative computers (such as the above) are more powerful than classical computers. This annoys me.
I told him I had a thesis that no physical system (except possibly for quantum computers) is more powerful than a Turing Machine, in terms of complexity.
Then he stated some version of the "Strong Physical Church-Turing Thesis":all computations done using a polynomial amount of matter and a polynomial amount of time can also be done with a Turing Machine with a polynomial-sized tape in polynomial time. (paraphrased)
I strongly believe that this holds for systems that don't use quantum entanglement (chemical computers, brains, etc don't).
UPDATE: I forgot to say: apparently, primality testing is in P!
|
|
| thoughts on Voronoi diagrams, and its dual |
[19 Jun 2009|09:42am] |
* In 2D, the dual of a Voronoi diagram is defined by linking cells that share an edge. In 3D, we can still draw these lines, but will they form cells?
* What is the name of the diagram defined by the intersections between Voronoi cells and dual cells?
* There's a trivial O(n^3) algorithm for computing whether two cells share an edge / whether an edge is present in the dual graph. I'm sure you can do better with kd-trees. But can you do even better?
* One talks of smoothing a signal (e.g. through a low-pass filter). I haven't yet heard of smoothing a sample. But if we take each sample point and have it follow the gradient of its cell's area, we should get a smoother version of the sample. This idea makes me feel good, because it's non-parametric, at least on the surface.
It's not totally analogous to smoothing because:
* the outer points don't move, since their area is already infinite. So in the limit you will get a roughly uniform distribution that is concentrated... with signal smoothing, the limit you spreading yourself thin: the density converges to the 0 function.
* if you do it for more than 1 iteration, you get some non-trivial dynamics, as the points consider their neighbors' changing location.
* the idea of using Voronoi diagrams may not seem smooth enough: you consider your immediate neighbors while ignoring those further away. One idea is to simultaneously consider different metrics (with different weights). You could e.g. weight the horizontal direction more or less according to a parameter a: d^2 = a*x^2 + y^2. I would like to see a data structure for a whole range of a: it would let you query efficiently, and the smoothing operation would consider several different diagrams: you're now following the gradient of the expected change in area.
I don't really have applications for these ideas (other than two-sample tests). It's easy to go overboard with mathematical mind candy like this.
|
|
| randomness-deficiency and game theory |
[19 Jun 2009|01:08am] |
Randomness is a finite resource. People seem to have less of it than computers. (I'd say this is a form of "bounded rationality")
What does this imply about game theory, namely mixed strategies?
UPDATE: do correlated equilibria extend classical game theory in the required way?
|
|
| linguistic oddity |
[14 Jun 2009|10:52am] |
For the past year, I've been noticing a new phenomenon among English speakers: when pausing and thinking anxiously, they fill the pause by saying ch-chch-chch-ch... it's like they are singing a beat.
I interpret it as saying: "I don't wish to be interrupted, and will come back to you in just a moment", and can be used while thinking or while driving.
I have observed this in at least 25/M/Canadian, 30/F/Canadian, but I suspect it's widespread in the USA too.
Is this a new phenomenon?
|
|
| multivariate two-sample tests |
[12 Jun 2009|03:33pm] |
A summer school colleague who is a neuroscientist recently asked for a way to test whether distributions in 4D space are the same.
First, I came up with the idea of using Voronoi cells as bins: let sample 1 define the Voronoi cells (bins), and points from sample 2 will fall into them. If they come from the same distribution, the set of bins will have a uniform distribution. (Computing how unlikely a deviation is isn't that easy: if sample 1 were infinite, the bins would have a binomial distribution same with the same parameter, but what should you do in reality?)
To prove my claim of uniformity, we need the fact that, if we're sampling IID from a joint random variable, then there exists a constant k such that for any point p, the area of the Voronoi cell around p converges in probability to k/n(density at p), as n -> infty. (I think that k=1) [Did it state this too strongly?]
Then I did a brief literature review, which I quite enjoyed doing because all these ideas are so intuitive.
In some of these tests, you can flip around the two samples, and run the test again. In such cases, the p-value should ideally be defined by a combination of the two scores (it seems a priori plausible that one direction rejects, while the other direction fails to).
|
|
| kids these days... |
[10 Jun 2009|07:40pm] |
|
Since when is Facebook the standard way of emailing someone?
|
|
| Santa Fe; ToDo list |
[08 Jun 2009|07:42pm] |
This city of Santa Fe really has a weird architecture, like everything is made of clay. Reminded me of DisneyWorld. We're in the suburbs though, but only a ~20-minute walk from downtown.
I've met tons of people so far (I'm also recording who and when), and try to study their names.
---
ToDo:
* detergent, toothpaste DONE
* nasal gel (Ayr), sunscreen (I felt my nose dry upon arrival... and my skin burning after the reception yesterday)
* buy GSM card, calling card
* get a bicycle
* improvise a fridge (or a styrofoam substitute), since cafeteria dinner closes at 6:30pm
* buy beard clipper insert, which broke inside the luggage
---
UFO: there was a weird light on the sky a couple hours ago. Like Venus, but much much brighter. I wondered if it was the moon, but it was way too small, and not quite round. I took pictures. Hmm... Roswell isn't too far from here.
---
Geoffrey West on SFI funding:
* SFI does not accept too much money from any one source (including government), as a way to prevent grant-slavery. One NSF proposal submitted by SFI had 32 reviewers (maybe because they didn't know what to do with it?)
* If you're a private donor, SFI is a cheap way to buy impact (100 times more efficient than Harvard). The institute's total impact factor seems to be comparable to Penn, despite having much less money.
---
My housemates are talking about satisficing, the math-fetish of some economists, simulations of emotions, and how political activism seems to be genetically inheritable.
---
Unlike the IPAM Summer School, few people here speak the same language as me. My impression is that there are lots of ecologists and political scientists, most of whom seem to know their math.
During introductions today, most people mentioned extra-curricular activities. Soccer, frisbee, dancing, hiking, biking. I pitched the Telluride Bluegrass Festival, which I'm planning to attend with my guitar-playing housemate.
|
|